180 BOTANICAL GAZETTE [SEPTEMBER 
In 2 days its weight was 59 mg. Fig. 4 shows the curve for loss 
of water from such a seed which weighed 198.5 mg. at the time it 
was put into NaCl. 
The middle layer, which is several cells thick, was subjected to 
a direct test by using it as an osmotic membrane cemented over 
a rubber cork. Its osmotic character is demonstrated by the 
following figures: 
Rise Time Rise Time 
5 We a oe 5 hrs $55 ON, savin ane 32 hes 
“cc iff 
OP a ee Sees oy 9 £00 ee (oe ee 8 
“ “ ” ‘“c “ 
0G es es 20 by F Pea er gee een gn 73 
However, this layer is by no means as efficient as the inner layer. 
The NaCl could be seen streaming through the membrane into the 
2 distilled water, causing the same appearance as the rapid 
solution of a salt in water. 
Neither the middle nor inner membrane is as perfectly 
semipermeable alone as are both together in the natural 
condition. Microchemical study of the 
coat, using NaCl and AgNO3, had given 
CELLET LOTT SEY evidence in the same direction, that the — 
Fic. 4—Curve showing Middle layer allowed salts to reach the 
loss of water by a seed inner layer, but that the precipitate of 
which had become excessively AgC] was not certainly present within the 
eee inner layer. The extreme stretching of 
outer coats of the testa had the coats preceding the bursting, or 4 
burst, leaving the inner coat, physical or chemical modification due to 
one cell thick, intact; the the macerating agent, may be respon- 
curve demonstrates the semi- 
permeable character of the Sible for the impaired eee of the — 
inner layer. separated membranes. 
The cellular arrangement in the inner 
membrane is shown in fig. 5. Microchemical tests showed the 
walls of this layer to be largely cellulose, with little or no suberiza-_ 
tion. A dark blue color was obtained on treatment with sulphuric 
acid and iodine, and the materials of which it is composed are 
almost completely dissolved in Schweizer’s reagent. Practically 
no coloration was obtained with alkanin and other suberin reagents, 
except in the cell contents, and these are known not to be respon- 
~ 
g 

s 
i eeonre ers. eee ee 

° 



